Research On Application Of Multi-core SVM Algorithm In Android Malware Detection

In recent years,the rapid development of mobile Internet lead to the mobile terminal equipment being more and more indispensable in people's work and life.On half of the operation system,Android mobile phone has quickly occupied the market.It brings a convenient and simple life,however,the security problems emerge in an endless stream.Malware is becoming more and more rampant,and it is necessary to improve the detection and protection of protective software.It is also imperative to study the detection of malicious applications.Malicious software detection is essentially a two classification problem and the ultimate goal is to detect whether an application is malicious.Based on this,this paper chooses to use a very wide support vector machine algorithm to solve the problem on the classification problem.The algorithm is improved from two aspects: kernel function and parameter optimization,and the improved support vector machine algorithm is applied to the Android malware detection model.Therefore,the main contents of this paper include:(1)In order to improve the generalized ability of Gaussian kernel function,the Gaussian kernel function is improved by introducing new control parameters,after comparison with the original Gaussian kernel function,which not only improves the generalization ability but also has strong local learning ability.(2)In view of the strong local learning ability of Gaussian kernel function and the generalization ability of polynomial kernel function,I propose a combined kernel function combining the two and construct a new support vector machine algorithm with this kernel.In the light of the choice of parameters in the support vector machine algorithm,this paper analyzes and compares several commonly used parameter optimization algorithms,and proposes an improved particle swarm optimization algorithm to optimize the parameters to ensure that the final classification result is achieved the best.(3)An Android malware detection model is constructed.The algorithm is applied to the detection model.The experimental results show that the proposed algorithm can detect malware well.